Increases in the prevalence and duration of drought events due to climate change are a major concern agricultural production faces, with potentially destabilizing consequences for food security and sustainability at local through to global scales [1–3]. As a result, managing and predicting crop yield under drought is among the most pressing goals for agricultural management under environmental change, and has been for decades [4–9]. In aiming to understand the mechanistic basis of yield change under drought, one primary theme in crop science has been to better understand how plants cope with periods of water stress [10, 11].
Several reviews exist that summarize the myriad of short- and long-term drought tolerance mechanisms and strategies in plants globally, which spans from short-term physiological responses such as biochemical signaling that triggers stomatal closure, through to longer-term responses including changes in leaf-, root-, or whole-plant traits [2, 3, 12–15]. Indeed, this deep literature demonstrates how and why drought tolerance represents among the most complex and multifaceted plant characteristics, and is dictated by the cumulative expression of multiplebiochemical, anatomical, architectural, and morphological traits [13]. Owing to its complex nature, studies often use different approaches or metrics to quantify drought tolerance. These metrics utilize various functional traits including instantaneous water-use efficiency (measured as a ratio of photosynthetic carbon gain [A] to water loss through transpiration [E] [e.g., 16]), stomatal traits including conductance (gs) and sensitivity [e.g., 17, 18], traits associated with root phenotypes [e.g., 19], among others [e.g., 20].
One trait employed to characterize drought tolerance in plants is the turgor loss point (πtlp), which represents the leaf water potential (Ψleaf) at which wilting occurs [21–23]. This trait is widely considered an indicator of drought tolerance, as plants expressing a more negative πtlp are able to maintain leaf turgor and physiological functioning including sustained rates of gs, A, and E, across a wider range of moisture availability [23]. In the field of comparative plant and functional trait ecology, turgor loss point is considered a “higher-level” drought tolerance trait that can be used to infer leaf- and plant-level environmental tolerances [22]. As such, πtlp and its plasticity have been employed to characterize drought tolerance across large groups of species [21, 22, 24, 25]. In turn, in studies on unmanaged ecosystems, inter- and intraspecific variation in πtlp is used to support hypotheses surrounding the environmental determinants of plant species distributions [26, 27] and species and community-level responses to changing water availability [e.g., 18, 24, 28].
More recently, and largely as an extension of the literature on “wild” plants, πtlp has received attention as a hypothesized index of drought tolerance in crops. For example, in detecting variation in πtlp within and among eight cultivars of spring wheat (Triticum aestivum L.), Mart et al. [29] argued that πtlp represents a trait that could be used to rapidly screen crop drought tolerance. Employing πtlp as a crop drought tolerance trait is relatively new, since prior to this time, there was a widespread assumption that crop physiological functioning ceased below a certain soil-based permanent wilting point [as cited by 29, 30]. Furthermore, researchers have noted that relationships between πtlp and crop growth and yield remain an unresolved area of research, leading to uncertainties as to whether or not this trait represents a viable screening tool in breeding programs that are relevant for agricultural management [31]. Nonetheless, because πtlp is a metric that integrates multiple aspects of osmotic adjustment in plants, quantifying πtlp within and among crop species and varieties appears an increasingly important component of the wider literature on quantifying crop drought tolerance.
To date pressure-volume (P-V) curves have been the classical method for estimating πtlp in plants, species, or genotypes [reviewed by 32], with P-V curve-derived data then supporting literature focused on plant drought tolerance and its ecological implications [33]. A P-V curve is constructed by progressively drying a fully rehydrated leaf at set pressure or drying intervals, and then assessing the statistical relationship between Ψleaf and relative water deficit (RWD) [summarized in Fig. 1 of 22]. Data necessary for P-V curve construction and associated πtlp estimation are generated with the use of Scholander-type pressure chambers or “pressure bombs” [34], generally following either a bench drying method where leaves are air dried on a bench top and both Ψleaf and RWD are measured at set time intervals, or a “squeeze method” where the mass of expressed sap is weighed at set pressure intervals.
Despite being widely applied in both basic and applied plant sciences for decades [34, 35], P-V curve generation is associated with major time constraints [33], often taking hours to complete for a single leaf, and certain assumptions embedded within P-V curve-based methods have motivated critical reviews on their accuracy across studies [32, 33, 36]. For example, Rodriguez-Dominguez et al. (2022) elucidated how key assumptions related to sample preparation, storage, and pressurization techniques (among other factors) may lead to variability in Ψleaf measurements, and ultimately P-V curve construction from pressure chambers. At the same time, estimating πtlp for a single leaf through a P-V curve can take hours, even when using the relatively fast squeeze method [36]. So, while P-V curves remain a critical aspect of plant and crop science, there are major limitations associated with this technique for either 1) screening drought tolerance across multiple crops and varieties, or 2) for assessing variation in drought tolerance traits at the individual plant level. The latter is especially important in the fields of crop science and agroecology, where intraspecific trait variation—i.e., variation existing below the species level—is likely a key determinant of agroecosystem processes [37].
In response to these limitations, researchers have begun developing methods for high-throughput estimation of Ψleaf and πtlp, which have gained considerable interest in studies of woody and herbaceous species in unmanaged ecosystems [33, 36, 38–40], and more recently of crops [29]. Notably, studies have employed both vapour pressure osmometers [33, 38, 39] and—the focus of our study here—dewpoint hygrometers [36, 40] to estimate πtlp in plants by estimating this trait directly as a function of Ψs at full turgor. Specifically, a dewpoint hygrometer measures the sum of matric potential (Ψm) and osmotic potential (Ψs) using the chilled-mirror dewpoint technique, while the vapour-pressure osmometer measures osmolality which is then converted into Ψs following the Van't Hoff equation [33, 36]. In both these methods gravitational potential (Ψg), matric potential (Ψm), and Ψp are considered negligible when leaves are fully hydrated, and as a result Ψs in fully hydrated leaf samples can then be correlated to πtlp [33].
The use of dewpoint hygrometers for measuring leaf bulk water relations dates back decades [41, 42] and recent studies using these techniques in high-throughput assessments of plant ecophysiology are promising. Specifically, Petruzellis et al. [40] found this method could capture interspecific variation in πtlp across 27 Mediterranean woody species (which ranged from ~-4.5 to -0.5 MPa), with dewpoint hygrometer-based measurements of Ψs at full turgor linearly predicting (adjusted r2 = 0.46) πtlp values derived from P-V curves. Within species, Banks and Hirons [36] found this technique was able to quantify fine-scale differences in πtlp that exists among five Acer genotypes, with mean πtlp ranging between roughly − 1.5 to -2.0 MPa: differences that were masked when πtlp was measured using P-V curves alone. While these studies and foundational theory [41, 42] point to the dewpoint hygrometer as a viable technique for high-throughput πtlp estimation, no studies have yet tested if this technique is able to quantify variation in πtlp across crop varieties.
Wine grapes (Vitis vinifera subsp. vinifera) represent among the world’s most economically important crops, with the environmental conditions necessary for wine grape production varying widely across the world’s over 6,000 varieties [43–46]. Many varieties require a narrow range of climatic conditions for optimum plant performance, physiological functioning, fruit quality, and yield [45, 46]. Indeed, authors have argued that wine grapes are among the most sensitive to climatic shifts with projections indicating that due to alterations in water availability and temperature regimes, wine production will likely shift considerably in the future [47, 48]. Wine grape drought tolerance is an integrated characteristic comprised of multiple genetic, morphological, physiological, and phenological traits [49]. Additionally, the viability of wine grape production under a shifting climate is mediated not just by drought tolerarnce, but also by other agronomic considerations such as berry size, yield, and flavor profiles. No less, high-throughput measurements of πtlp would be valuable towards holistic prediction and modelling of climatic suitability of varieties into the future [47, 50].
In this study, we assess whether or not a high-throughput technique based on the use of a dewpoint hygrometer, is able to quantify variation in πtlp both within and among wine grape varieties. To address this, we specifically compare differences in πtlp generated through P-V curves and a dewpoint hygrometer, at both the individual leaf- and variety-scale. While the primary focus of our study is on πtlp variation within and among 12 widely cultivated varieties of wine grapes, our study also includes measurements on a wild relative of wine grapes, namely Vitis riparia, in order to assess the wider application of this technique towards quantifying πtlp in wild crop relatives [51]. Our study was designed to address the following research questions: 1) do varieties of wine grapes vary significantly in their πtlp, and if so, can these differences be quantified using a high-throughput technique? Then, we asked 2) which varieties are most drought-tolerant as per πtlp values, and if inferences regarding varietal drought tolerance rankings change depending on πtlp methods? Finally, we ask 3) do cultivated wine grapes differ in their πtlp vs. wild relatives?